Add constant write backoff in GelfTCPSender for cases when a sender c… #249
Conversation
…annot write any bytes to the socket channel several times. Limit the time to stay in such mode.
| } catch (InterruptedException e) { | ||
| Thread.currentThread().interrupt(); | ||
| } | ||
| long sleepTimeMs = TimeUnit.NANOSECONDS.toMillis(System.nanoTime() - startTime); |
There was a problem hiding this comment.
Nano time is prone for overflows. Millis should be fine.
There was a problem hiding this comment.
could you please share more details (maybe some link to the description) regarding this issue?
it is a common pattern of nanotime usage..
Regarding overflow I am aware only about this point:
https://docs.oracle.com/javase/8/docs/api/java/lang/System.html#nanoTime--
Differences in successive calls that span greater than approximately 292 years (2^63 nanoseconds) will not correctly compute elapsed time due to numerical overflow.
System.currentTimeMillis() may have not enough precision in some cases.
src/main/java/biz/paluch/logging/gelf/intern/sender/DefaultGelfSenderProvider.java
Show resolved
Hide resolved
src/main/java/biz/paluch/logging/gelf/intern/sender/GelfTCPSender.java
Outdated
Show resolved
Hide resolved
…new parameters to docs, stop attempts to send a message in case of InterruptedException
for cases when a sender cannot write any bytes to the socket channel several times. Limit the time to stay in such mode. stop attempts to send a message in case of InterruptedException. Original pull request: #249
|
Thank you for your contribution. That's merged and polished now. |
|
Thank you a lot for the quick responses. |
|
1.14.1 is available from Maven Central. I decided to ship a release earlier this morning since the last one was shipped about 6 months ago. |
…annot write any bytes to the socket channel several times. Limit the time to stay in such mode.
fixes #248